Regression Cases

Maintain public, non-sensitive hashes for each supported case and rerun them after workflow or script changes. Keep expected assertions factual and minimal so tests do not rot with labels or presentation changes.

Standard Cases

• Successful native transfer with no logs.
• Successful ERC-20 transfer.
• ERC-721 and ERC-1155 transfers.
• Approval without an asset transfer.
• Router swap with callback and fee.
• Wrap and unwrap transaction.
• Contract deployment.
• Verified proxy using DELEGATECALL.
• Proxy upgrade with confirmed implementation change.
• Batch or multicall with several effects.
• Account-abstraction transaction.
• Bridge source transaction with no assumed destination result.

Failure Cases

• Error(string) revert.
• Panic(uint256) revert.
• Verified custom error.
• Empty revert data.
• Failed child propagated to the top.
• Failed child caught by a successful parent.
• Out-of-gas or exceptional halt supported by trace evidence.

Assertions

For every case, check:

• Correct chain and receipt status.
• Exact raw values, deterministic execution gas fee, and conservative total-fee availability when chain-specific components are missing.
• No invented labels, names, decimals, or errors.
• No complete-call-tree claim in Standard mode.
• Committed and attempted effects remain distinct.
• Proxy and implementation roles are correct.
• Explorer links use the right host.
• Confidence matches evidence depth.

Do not embed API keys, authenticated RPC URLs, private transactions, or personal user data in regression fixtures.